t/SeqIO/table.t

   1 # -*-Perl-*- Test Harness script for Bioperl
   2 # $Id$
   3
   4 use strict;
   5
   6 BEGIN {
   7     use lib '.';
   8     use Bio::Root::Test;
   9
  10     test_begin(-tests => 450,
  11                            -requires_module => 'IO::Scalar');
  12
  13         use_ok('Bio::Tools::CodonTable');
  14         use_ok('Bio::SeqIO::table');
  15 }
  16
  17 my @names = qw(A6
  18                A6r
  19                A6ps1
  20                A6ps2
  21                CaMK2d
  22                CaMKK2
  23                AMPKa1
  24                AMPKa2
  25                MARK3
  26                MARK2);
  27 my @accs = qw(SK001
  28               SK512
  29               SK752
  30               SK766
  31               SK703
  32               SK482
  33               SK032
  34               SK033
  35               SK096
  36               SK120);
  37 my @num_anns = (5, 5, 5, 5, 6, 7, 7, 7, 7, 7);
  38 my @psg = (0, 0, 1, 1, 0, 0, 0, 0, 0, 0);
  39 my @rs = (0, 0, 0, 0, 1, 1, 1, 1, 1, 1);
  40
  41 ok my $seqin = Bio::SeqIO->new(-file => test_input_file("test.tsv"),
  42                             -format  => 'table',
  43                             -species => "Homo sapiens",
  44                             -delim   => "\t",
  45                             -header  => 1,
  46                             -display_id => 1,
  47                             -accession_number => 2,
  48                             -seq => 7,
  49                             -annotation => 1,
  50                             -trim => 1);
  51 run_tests([@names],[@accs],[@num_anns],[@psg],[@rs]);
  52
  53 $seqin->close();
  54
  55 ok $seqin = Bio::SeqIO->new(-file => test_input_file("test.tsv"),
  56                          -format  => 'table',
  57                          -species => "Homo sapiens",
  58                          -delim   => "\t",
  59                          -header  => 1,
  60                          -display_id => 1,
  61                          -accession_number => 2,
  62                          -seq => 7,
  63                          -colnames => "[Family,Subfamily,Pseudogene?,Protein,Novelty]",
  64                          -trim => 1);
  65 run_tests([@names],[@accs],[4,4,4,4,4,5,5,5,5,5],[@psg],[@rs]);
  66
  67 $seqin->close();
  68
  69 ok $seqin = Bio::SeqIO->new(-file => test_input_file("test.tsv"),
  70                          -format  => 'table',
  71                          -species => "Homo sapiens",
  72                          -delim   => "\t",
  73                          -header  => 1,
  74                          -display_id => 1,
  75                          -accession_number => 2,
  76                          -seq => 7,
  77                          -annotation => "[4,5,6,8,10]",
  78                          -trim => 1);
  79 run_tests([@names],[@accs],[4,4,4,4,4,5,5,5,5,5],[@psg],[@rs]);
  80
  81 $seqin->close();
  82
  83 # need Spreadsheet::ParseExcel installed for testing Excel format
  84 SKIP: {
  85         test_skip(-tests => 112, -requires_module => 'Spreadsheet::ParseExcel');
  86
  87         ok $seqin = Bio::SeqIO->new(-file => test_input_file("test.xls"),
  88                                                          -format  => 'excel',
  89                                                          -species => "Homo sapiens",
  90                                                          -header  => 1,
  91                                                          -display_id => 1,
  92                                                          -accession_number => 2,
  93                                                          -seq => 7,
  94                                                          -annotation => 1,
  95                                                          -trim => 1);
  96         run_tests([@names],[@accs],[@num_anns],[@psg],[@rs]);
  97
  98         $seqin->close();
  99 }
 100
 101 sub run_tests {
 102     my ($names_,$accs_,$num_anns_,$psg_,$rs_) = @_;
 103
 104     my @names = @$names_;
 105     my @accs = @$accs_;
 106     my @num_anns = @$num_anns_;
 107     my @psg = @$psg_;
 108     my @rs = @$rs_;
 109
 110     my $n = 0;
 111     my $translator = Bio::Tools::CodonTable->new(-id => 1);
 112     while (my $seq = $seqin->next_seq()) {
 113         $n++;
 114         is ($seq->display_id, shift(@names));
 115         is ($seq->accession_number, shift(@accs));
 116         ok ($seq->species);
 117         is ($seq->species->binomial, "Homo sapiens");
 118         my @anns = $seq->annotation->get_Annotations();
 119         is (scalar(@anns), shift(@num_anns));
 120         @anns = grep { $_->value eq "Y";
 121                      } $seq->annotation->get_Annotations("Pseudogene?");
 122         is (scalar(@anns), shift(@psg));
 123
 124         # check sequences and that they translate to what we expect
 125         if (($n >= 5) && ($seq->display_id ne "MARK3")) {
 126             my $dna = $seq->seq;
 127             my $protein = "";
 128             my $frame = 0;
 129             while ($frame <= 2) {
 130                 my $inframe = substr($dna,$frame);
 131                 # translate to protein
 132                 my $protseq = $translator->translate($inframe);
 133                 # chop off everything after the stop and before the first Met
 134                 while ($protseq =~ /(M[^\*]+)/g) {
 135                     $protein = $1 if length($1) > length($protein);
 136                 }
 137                 $frame++;
 138             }
 139             # retrieve expected result from annotation and compare
 140             my ($protann) = $seq->annotation->get_Annotations("Protein");
 141             ok (defined $protann);
 142             is ($protein, $protann->value);
 143         }
 144
 145         @anns = grep { $_->value eq "Known - Refseq";
 146                      } $seq->annotation->get_Annotations("Novelty");
 147         is (scalar(@anns), shift(@rs));
 148         @anns = $seq->annotation->get_Annotations("Subfamily");
 149         is (scalar(@anns), ($n <= 5) ? 0 : 1);
 150         @anns = $seq->annotation->get_Annotations("Family");
 151         is (scalar(@anns), 1);
 152         is (substr($anns[0]->value,0,4), ($n <= 4) ? "A6" : "CAMK");
 153     }
 154
 155     is ($n, 10);
 156 }